Recent studies have demonstrated that ictal SPECT (Single Positron Emission Computed Tomography) is a useful noninvasive method for localizing seizure foci in candidates for epilepsy surgery. Despite this, relatively few centers apply this technique routinely, due to the difficulty of ensuring optimal ictal injection and the lack of adequate software for the necessary image analysis. The overall goal of: t:his project is to develop software tools to facilitate the use of ictal SPECT for presurgical localization. Phase I will focus on the feasibility of implementing a patient-specific warning system capable of reliable prediction or early detection of seizure onset from scalp EEG, by developing trainable seizure detection software to detect or predict seizures in a given patient based on previous seizures. Ictal EEG dipole model in with localization constrained by SPECT foci will also be investigated, which may be a useful tool for checking concordance between ictal blood flow changes and the seizure discharge, and for assessing dynamic EEG relationships between multiple foci in rapidly spreading seizures. Phase II will focus on optimizing the seizure warning system and implementing realtime software for field testing, in addition to systematically evaluating the clinical value of ictal dipole modeling constrained by SPECT. PROPOSED COMMERCIAL APPLICATION: A successful outcome will substantially improve the clinical application of SPECT to locate the specific zone in the brain where seizures start and thus minimize the use of intracranial electrodes in presurgical diagnostic studies. This will significantly reduce the costs and patient stress associated with these studies.